A Study on the Apparent Randomness of a Wildlife Sample


  • Christine Kraamwinkel* University of Pretoria
  • Inger Fabris-Rotelli University of Pretoria




Sampling is used to estimate characteristics of the population when we are unable to investigate the population as a whole. In an ideal world a sample would be a perfectly scaled-down version of the original population in the sense that every characteristic of the population would be matched in the sample. Although this ideal is almost impossible to meet, researchers aim to get as close to this as possible. Even though wildlife researchers are aware of the advantages of random sampling, these methods are usually not implemented. In practice, most samples are convenience samples, so the selection probabilities of the elements cannot be described, making it impossible to derive statistically valid estimators and their errors. Typically, it is assumed that these convenience samples approximate random samples so that inferences can be made about the population, however, these assumptions remain mostly unfounded and untested. In wildlife research, probability sampling methods such as simple random sampling (SRS) are not practical since all elements in the population may not be available or accessible. Instead, prior knowledge is often used to select elements, or in some cases, any available element is included. Only a small number of studies on this aspect have been done.

This paper will assess the impact of taking a convenience sample by making use of cattle livestock data. In this study, a convenience sample was obtained by selecting 10\% of the farmers registered at each of the dip tanks.В  Census data is available for this population. We aim to provide measures of how the quality of the sample, in other words the randomness or nonrandomness of the sample, affects statistical analysis. We aim to show that a convenience sample obtained in this setting will yield less reliable results than a probability sample. We would like to add a measure attached to a convenience statistical analysis in order to make a comparison with the unknown statistical analysis attached to a true random sample.

Author Biographies

Christine Kraamwinkel*, University of Pretoria

Junior Lecturer, Department of Statistics

Inger Fabris-Rotelli, University of Pretoria

Senior Lecturer, Department of Statistics






Conference Contributions